Metrics

The Metrics section of the dashboard provides real-time charts that track your agent's performance over time.

Summary Row

At the top, a row of headline numbers gives you the quick picture: total tokens, active sessions, error count, and uptime percentage — all for the selected time range.

Token Usage Chart

A stacked area chart showing token consumption over time, broken down by model. Use the Model Selector to filter by specific models or view all at once.

This is the fastest way to spot cost spikes or unexpected model switches.

Errors & Warnings Chart

A bar chart counting errors and warnings per time bucket. Each bar is split by severity so you can see whether issues are critical or informational at a glance.

Uptime Chart

A horizontal strip chart showing gateway availability over the selected period. Green means healthy, gaps or red segments indicate downtime. Hover over any segment to see the exact timestamps.

Time Granularity

Use the granularity picker to adjust the time window:

Granularity	Best for
30m	Debugging a live issue
1h	Recent session review
6h	Morning/afternoon overview
12h	Half-day trend
24h	Daily summary

Real-Time Updates

All charts update automatically via GraphQL subscriptions — no manual refresh needed. When new data arrives, charts animate smoothly to include it.

TIP

Combine the metrics view with the Events log to correlate error spikes with specific events.